Optimization of Petlyuk sequences using a multi objective genetic algorithm with constraints

In this work we use genetic algorithms to optimize Petlyuk sequences using a rigorous design model. A multi objective genetic algorithm (GA) with constraints was formulated and interconnected with the Aspen Plus process simulator to obtain each data point during the search process. In addition to providing more energy-efficient designs than some reported structures, two relevant trends were observed from the results of the case studies; one had to do with the feed location to the prefractionator as a function of the mixture properties, and the other one with optimal structures requiring four interconnecting stages instead of the two normally used for Petlyuk sequences. An application for the separation of azeotropic mixtures is also included. The optimal placement of vapor–liquid interconnections is again shown to be different for each interconnecting stream. The GA showed a robust performance, and was practically independent on the initial values for the search variables.